Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 © Author(s) 2021. This work is distributed under the Creative Commons Attribution 4.0 License.

Zooplankton mortality effects on the plankton community of the northern Humboldt Current System: sensitivity of a regional biogeochemical model

Mariana Hill Cruz1, Iris Kriest1, Yonss Saranga José1, Rainer Kiko2, Helena Hauss1,3, and Andreas Oschlies1,3 1GEOMAR Helmholtz Centre for Ocean Research Kiel, Düsternbrooker Weg 20, 24105 Kiel, Germany 2Laboratoire d’Océanographie de Villefranche-sur-Mer, Sorbonne Université, Villefranche-sur-Mer, France 3Kiel University, 24098 Kiel, Germany

Correspondence: Mariana Hill Cruz ([email protected]) and Iris Kriest ([email protected])

Received: 11 November 2020 – Discussion started: 9 December 2020 Revised: 16 March 2021 – Accepted: 22 March 2021 – Published: 12 May 2021

Abstract. Small pelagic fish off the coast of Peru in the east- 1 Introduction ern tropical South Pacific (ETSP) support around 10 % of global fish catches. Their stocks fluctuate interannually due Eastern boundary upwelling systems (EBUSs) are among the to environmental variability which can be exacerbated by most productive regions in the ocean. Despite their small fishing pressure. Because these fish are planktivorous, any size, they support a large fraction of the world’s fisheries change in fish abundance may directly affect the plankton (Chavez and Messié, 2009). The northern Humboldt Cur- and the biogeochemical system. rent System (NHCS) in the eastern tropical South Pacific To investigate the potential effects of variability in small (ETSP) Ocean is the most productive EBUS, producing 10 % pelagic fish populations on lower trophic levels, we used a of global fish catches (Chavez et al., 2008) and supporting the coupled physical–biogeochemical model to build scenarios fishery of the Peruvian anchovy Engraulis ringens, which is for the ETSP and compare these against an already-published the biggest single-species fishery on the planet (Chavez et al., reference simulation. The scenarios mimic changes in fish 2003). The ETSP is also characterised by substantial interan- predation by either increasing or decreasing mortality of the nual variability (i.e. El Niño–Southern Oscillation; Holbrook model’s large and small zooplankton compartments. et al., 2012) and an intense midwater oxygen minimum zone The results revealed that large zooplankton was the main (OMZ), resulting in high denitrification rates (Farías et al., driver of the response of the community. Its concentration in- 2009). creased under low mortality conditions, and its prey, small As in other EBUSs, small pelagic fish are highly abundant zooplankton and large phytoplankton, decreased. The re- (Cury et al., 2000) in the NHCS, building up large popula- sponse was opposite, but weaker, in the high mortality sce- tions that are severely affected by climate fluctuations. For narios. This asymmetric behaviour can be explained by the example, anchovy biomass in the NHCS fluctuated between different ecological roles of large, omnivorous zooplankton 10 × 106 and 16 × 106 t in the 1960s (Alheit and Niquen, and small zooplankton, which in the model is strictly her- 2004). Its area of distribution spans from northern Peru to bivorous. The response of small zooplankton depended on northern Chile and the Talcahuano region off central Chile the antagonistic effects of mortality changes as well as on (Fig. 1 in Alheit and Niquen, 2004). During the El Niño event the grazing pressure by large zooplankton. The results of this of 1972, it dropped to 6 × 106 t (Alheit and Niquen, 2004), study provide a first insight into how the plankton ecosystem presumably caused by warming and the resulting decrease in might respond if variations in fish populations were modelled upwelling and production. These unfavourable growth con- explicitly. ditions for anchovy might have been exacerbated by fishing pressure (Beddington and May, 1977; Hsieh et al., 2006). From 1992 to 2008, the population of anchovy off the Peru-

Published by Copernicus Publications on behalf of the European Geosciences Union. 2892 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS vian coast fluctuated between 3×106 and 12×106 t (Fig. 13 levels. Zooplankton mortality may also form the link to fish in Oliveros-Ramos et al., 2017). models, when these are explicitly considered in the context The Peruvian anchovy is a planktivorous fish whose diet of biogeochemical models (e.g. Travers-Trolet et al., 2014b). changes over its ontogenic development. Anchovy first- However, there is no consensus on the form of the mortal- feeding larvae consume mainly phytoplankton, and when ity term, linear (µ·[Z], where [Z] is the zooplankton concen- they reach a length of 4 mm their diet gradually switches tration and µ is a mortality rate) and quadratic (µ·[Z]2) being to zooplankton, especially nauplii of copepods (Muck et al., two common forms (e.g. Evans and Parslow, 1985; Fasham 1989). Adult anchovies’ main sources of energy are eu- et al., 1990; Koné et al., 2005; Kishi et al., 2007; Aumont phausiids and copepods although phytoplankton is still found et al., 2015). A common argument for preferring quadratic in their diet (Espinoza and Bertrand, 2008). The other promi- to linear mortality is the reduction in unforced short-term os- nent small planktivorous fish species in the NHCS is the Pa- cillations (Steele and Henderson, 1992), although Edwards cific sardine Sardinops sagax, which feeds on smaller parti- and Yool(2000) argue that quadratic mortality does not al- cles than anchovy, including phytoplankton and small zoo- ways remove such oscillations. A quadratic mortality term plankton (Ayón et al., 2008a). These two species can there- may also be interpreted as an increase in diseases because fore be expected to impose a direct top-down control on of high population densities, cannibalism or increased preda- plankton; at the same time, they may be bottom-up affected tion due to high densities of prey. Because it is very difficult by changes in plankton abundance caused by variations in to determine zooplankton mortality in the field, there is also physical forcing. no agreement on the exact value of mortality (either linear or Pauly et al.(1989) estimated that the total population of quadratic), and this term, in practice, is often adjusted to tune anchovy off Peru consumes 12.1 times its own biomass in the model. However, not using mortality rates based on ob- 1 year. Assuming an area of 6 × 1010 m2 (Ryther, 1969) and servations may limit the capability of the model to accurately a conversion factor of zooplankton wet weight to nitrogen represent the zooplankton compartment (Daewel et al., 2014) of 1000 mgww(mmolN)−1 (Travers-Trolet et al., 2014a), a and to draw predictions about the state and dynamics of the fluctuation in anchovy population of 9 × 106 t would result marine ecosystem (Anderson et al., 2010). Hirst and Kiørboe in a change in zooplankton mortality of 5 mmolNm−2 d−1 (2002) predicted a global mortality of copepods of 0.062 d−1 from anchovy predation alone, in a top-down-driven ecosys- at 5 ◦C and 0.19 d−1 at 25 ◦C in the field. Two-thirds of such tem assuming no non-linearities. The assumption that an- mortality is due to predation. In models, the values of the chovy can exacerbate a top-down control on zooplankton is quadratic mortality rate (hereafter called µZ) in the literature supported by a decline in zooplankton concentration in dense vary over a large range, from 0.025 (Fennel et al., 2006) up aggregations of anchovies (Ayón et al., 2008a,b). On the to 0.25 (mmolNm−3)−1 d−1 (Lima and Doney, 2004). other hand, co-occurring long-term fluctuations of zooplank- For the NHCS, the high variability in forcing, biogeo- ton and anchovies at the population scale also indicate a rel- chemistry and plankton and the high abundance of plank- evant bottom-up control in the NHCS (Alheit and Niquen, tivorous fish, together with its economic importance, indi- 2004; Ayón et al., 2008b). cate the need for end-to-end models that include details of Numerical models are valuable tools to examine the po- all components. However, developing such a model is chal- tential tight coupling across a large range of trophic lev- lenging, and studies in this area have so far focused on ei- els and the mutual interactions among the different compo- ther fish (Oliveros-Ramos et al., 2017) or physics and bio- nents, including top-down and bottom-up effects. Rose et al. geochemistry (Jose et al., 2017). Given the large importance (2010) pointed out the increasing need for so-called end- of zooplankton mortality as a link between these two model to-end models of the marine food webs; these set-ups cou- systems (Mitra et al., 2014) and the uncertainty associated ple models including physical and biogeochemical processes with it, in this study we focus on the effects of this parame- with models for higher trophic levels. When lower-trophic- ter on the biogeochemical system of the ETSP as a first step level (biogeochemical) and higher-trophic-level (fish) mod- towards a fully coupled system. els are coupled, the link is typically made at the plankton In order to model the highly dynamic nature of both physi- level, with the former models providing food for the latter cal and biogeochemical processes in the ETSP, we employed (e.g. Travers-Trolet et al., 2014b). a biogeochemical model specifically designed for EBUSs, In stand-alone biogeochemical models that do not include coupled to a finely resolved regional circulation model. The higher trophic levels, zooplankton mortality is a closure term, coupled model has already been validated against oxygen, used to return the additional biomass to detritus. It repre- nitrate and chlorophyll (Jose et al., 2017); this configuration sents all processes that reduce the concentration of zoo- serves as a starting point and reference for the sensitivity ex- plankton and are not explicitly included in the model (for periments. In this study, we focus on the four plankton groups instance, predation by gelatinous organisms, predation by of the model: small and large phytoplankton and small and higher trophic levels and non-consumptive mortality). For large zooplankton. Large phytoplankton is highly abundant example, Getzlaff and Oschlies(2017) used it to mimic pre- near the coast of Peru in the nutrient-rich upwelling region. dation and immediate egestion or mortality by higher trophic Towards the open ocean it is grazed down by its preda-

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2893 tor, large zooplankton; and further offshore, as conditions become oligotrophic, small zooplankton and phytoplankton take over. In this study, we first extend the model validation by Jose et al.(2017) and assess the reliability of the large zooplank- ton compartment in the model by comparing it against meso- zooplankton observations obtained by net hauls. Note that in this paper we always refer to simulated phyto- and zooplank- ton by their size class (small or large), while observations are referred to as mesozooplankton. We then present two sensi- tivity experiments in which we varied the mortality rate of quadratic zooplankton mortality by ±50%. Model sensitiv- ity is examined with regard to concentrations of model com- ponents and inter-compartmental fluxes. Besides the overall response of the prognostic variables to an increase or de- crease in zooplankton mortality, we describe how changing zooplankton mortality affects the trophic structure of plank- ton, with focus on the highly productive coastal domain. Fi- nally we discuss the implications of our study for the plank- ton community and for modelling higher trophic levels. Figure 1. Annually averaged sea surface temperature (◦C) and hor- izontal advection vectors (ms−1) and location of the analysed re- gions (see labels) and of a vertical section for analysing plankton 2 Methods spatial succession (white line at 12◦ S). The coastal upwelling re- gion (see Sect. 2.5) spans from the coast to about 40 to 50 km off- 2.1 ROMS–BioEBUS model set-up shore.

The Regional Oceanic Modeling System (ROMS; Shchep- etkin and McWilliams, 2005) is a high-resolution, free- of the coarse-resolution domain, the biogeochemical model surface, terrain-following coordinate ocean model that solves is forced with monthly nitrate and oxygen values from CARS the primitive equations considering Boussinesq and hydro- (Ridgway et al., 2002) and surface chlorophyll from SeaW- static assumptions. The Biogeochemical model for Eastern iFs (O’Reilly et al., 1998). Phytoplankton and zooplankton Boundary Upwelling Systems (BioEBUS), which was de- boundary conditions are derived from a vertical extrapola- tion of the chlorophyll data. Detailed information about the rived from a N2P2Z2D2 model from Koné et al.(2005), is coupled online to the physical part (Gutknecht et al., boundary and initial conditions and validation of the model 2013a,b). is available in Jose et al.(2017). In this study the coupled ROMS–BioEBUS model has 2.2 Biogeochemical model description the same configuration as in Jose et al.(2017). It contains a small, high-resolution domain forced by a larger coarse- The evolution of a biological tracer in time is represented resolution domain, using the AGRIF (adaptive grid refine- by Eq. (1). On the right-hand side of the equation, the first ment in Fortran) 2-way nesting procedure. The small inner ◦ term represents the advection with the velocity vector u. The grid has a horizontal resolution of 1/12 spanning from about eddy and molecular diffusion is represented by the second 69 to 102◦ W and from 5◦ N to 31◦ S (Fig.1). The large outer and third terms, where Kh is the horizontal diffusion coeffi- grid spans from 69 to 120◦ W and from 18◦ N to 40◦ S with ◦ cient and Kz the vertical diffusion coefficient. The last term is a resolution of 1/4 . The biological processes occur in three a source-minus-sink (SMS) term due to biological processes. time steps of 900 s for each physical time step of 2700 s in The full set of equations and detailed explanation about each both domains. The two domains have 32 sigma layers with process are available in Gutknecht et al.(2013a). a vertical resolution of less than 5 m at the surface and de- creasing to around 500 m at a maximum depth of 4500 m. ∂C ∂  ∂C  i = −∇·(uC )+K ∇2C + K i +SMS(C ) (1) The coarse-resolution domain temperature, salinity and ∂t i h i ∂z z ∂z i horizontal velocity are forced at the lateral boundaries with monthly climatological (1990–2010) SODA reanalysis (Car- The BioEBUS model is adapted for the biogeochemical ton and Giese, 2008). Both domains are forced at the surface processes specific to the low-oxygen conditions of EBUSs, with 1/4◦ resolution wind velocity fields from QuikSCAT with some processes being oxygen dependent (see Gutknecht (Liu et al., 1998) and monthly heat and freshwater fluxes et al., 2013a). It has two compartments of phytoplankton, from COADS (Worley et al., 2005). At the lateral boundaries two compartments of zooplankton and two compartments of https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2894 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS detritus. The zooplankton and phytoplankton groups are di- is the growth rate limited by light, temperature and nutrients: vided into two size classes (small and large). There is not an J max · αP · PAR explicit size parameter in the model; however, the compart- = Pi i JPi (PAR,T,N) q (4) ments aim at representing the main ecological functions of J max2 + (αP · PAR)2 Pi i the plankton community falling within each group. Hence, · f (NO−,NO−,NH+), small phytoplankton (PS) represents organisms smaller than Pi 3 2 4 about 20 µm that require low nutrients such as flagellates, − − + where f (NO ,NO ,NH ) is the growth limitation by nu- while large phytoplankton (P ) represents larger organisms Pi 3 2 4 L trients, PAR is the photosynthetically available radiation (see such as diatoms. Similarly, small zooplankton (Z ) simu- S Koné et al., 2005), J max is the maximal light-saturated lates the role of a zooplankton community smaller than about Pi growth rate which is a function of temperature and αP is 200 µm, such as ciliates, and large zooplankton (Z ) repre- i L the initial slope of the photosynthesis–irradiance (P–I) curve sents a community larger than 200 µm, such as copepods. (Gutknecht et al., 2013a). Large phytoplankton is charac- The two size classes of detritus (small D and large D ) are S L terised by a steeper initial slope of the P–I curve and by produced from phytoplankton and zooplankton mortality and larger half-saturation constants for nutrient uptake (see Ta- by release of unassimilated grazing material. The model also ble A1). Therefore, it grows better than small phytoplankton contains three compartments of dissolved inorganic nitrogen under low-light conditions, but its nutrient uptake increases (NH+, NO− and NO−), dissolved organic nitrogen (DON), 4 2 3 more slowly as nutrient concentrations increase. oxygen (O2) and nitrous oxide (N2O). The BioEBUS model (Gutknecht et al., 2013a) includes 2.2.2 Zooplankton oxygen-dependent remineralisation processes, which are di- vided into ammonification, nitrification and denitrification, Zooplankton increases its biomass through grazing on phy- as well as anammox, and are based on the formulations by toplankton and in the case of large zooplankton also on small Yakushev et al.(2007). N 2O is a diagnostic variable for zooplankton. Metabolism; mortality; and, in the case of small model output, and its production does not affect the concen- zooplankton, predation by large zooplankton are sink terms. tration of the other variables. It is based on the parameterisa- Predation by fish and other higher trophic levels is implicit in tion of Suntharalingam et al.(2000, 2012), which relates the the quadratic mortality term. The biomass lost by metabolism production of N2O to the consumption of O2 from decompo- and mortality is assumed to become detritus which may sink sition of organic matter in oxic and suboxic conditions. O2 to the sediments or become remineralised, and a small frac- concentrations depend on primary production, zooplankton tion of zooplankton losses become ammonium and dissolved respiration, nitrification and remineralisation. This model in- organic nitrogen, which is also subjected to remineralisation. cludes gas exchange of O2 and N2O with the atmosphere. The SMS terms of the small and large zooplankton com- partment are determined by Eqs. (5) and (6), respectively:

2.2.1 Phytoplankton SMS(Z ) = f 1 · (GPS + GPL ) ·[Z ] (5) S ZS ZS ZS S ZS 2 − G ·[ZL] − γZ ·[ZS] − µZ ·[ZS] , The SMS terms in the small and large phytoplankton com- ZL S S partments are determined by Eqs. (2) and (3), respectively: SMS(Z ) = f 1 · (GPS + GPL + GZS ) ·[Z ] (6) L ZL ZL ZL ZL L − ·[ ] − ·[ ]2 γZL ZL µZL ZL . = − · ·[ ] SMS(PS) (1 εPS ) JPS (PAR,T,N) PS (2) Here f 1ZS and f 1ZL are assimilation coefficients (see Xi PS PS also Table A1). G is feeding rates of predator Z (either − G ·[ZS] − G ·[ZL] − µP ·[PS], Zj j ZS ZL S large or small zooplankton) on prey X (small and large phy- = − · ·[ ] i SMS(PL) (1 εPL ) JPL (PAR,T,N) PL (3) toplankton and small zooplankton) calculated with the for- − PL ·[ ] − PL ·[ ] − ·[ ] mulation by Tian et al.(2000, 2001). There is a linear loss GZ ZS GZ ZL µPL PL S L rate accounting for basic metabolism (γ ) and a quadratic d Zi − wP · [P ]. loss rate also referred to as mortality. The mortality parame- L dz L ters µZS and µZL of the reference simulation are 0.025 and 0.05 (mmolNm−3)−1 d−1 for small and large zooplankton, respectively, as in Jose et al.(2017) and Gutknecht et al. Here GXi represents feeding rates by zooplankton (see Zj (2013a). ·[ ] Sect. 2.2.2); µPi Pi is the mortality term, representing The feeding rate follows the formulation from Tian et al. all not explicitly modelled phytoplankton losses; wPL is the (2000, 2001): sinking speed of large phytoplankton sedimentation, which e ·[X ] is an additional loss term for this compartment; εPi is the ex- Xi Zj Xi i G = gmax · , (7) Zj Zj + udation fraction of primary production; and JPi (PAR,T,N) kZj Ft

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2895

mesopelagic depths. Possible reasons for this mismatch will where g is the maximum grazing rate of predator Z , be discussed in Sect. 4.1. maxZj j eZ X is the preference of predator Zj for prey Xi, kZ is the j i j 2.4 Experimental design half-saturation constant and Ft is the total availability of food for predator Z . Large zooplankton responds more slowly j To mimic changes in grazing pressure on zooplankton due to changes in food due to the high k . In the case of large ZL to small-pelagic-fish biomass fluctuations, we followed the zooplankton F = e ·[P ]+e ·[P ]+e ·[Z ], and t ZLPS S ZLPL L ZLZS S approach by Getzlaff and Oschlies(2017) and varied the in the case of small zooplankton F = e ·[P ] + e · t ZSPS S ZSPL mortality rate of each zooplankton compartment by ±50 % [P ] (Gutknecht et al., 2013a). L in comparison to the reference scenario described by Jose 2.3 Zooplankton evaluation et al.(2017). Thereby, an increase in mortality assumes a large consumption of zooplankton by fish, while a decrease As noted above, this model was already validated against in mortality assumes fewer fish. Because the model does not oxygen, nitrate and chlorophyll (Jose et al., 2017). As a com- include an explicit compartment for fish, it is assumed that all plement, we here compare the large zooplankton compart- zooplankton biomass consumed by fish becomes part of the ment of the model, averaged from January to March, with detritus pool via immediate fish mortality and defecation. In observational data collected on RV Meteor cruise M93. The reality, a fraction of the biomass is extracted from the system samples obtained during this cruise include day- and night- by the fishing industry, predation by sea birds that defecate time hauls with a Hydro-Bios multinet (nine nets, 333 µm over land and migrations. mesh) between 10 February and 3 March 2013 on a tran- Our model has two zooplankton compartments. In order to sect off the Peruvian coast (≈ 12◦ S; see Fig.2f), captur- explore the different roles of large zooplankton as top preda- ing the vertical and horizontal gradient in zooplankton con- tor and small zooplankton as grazer and prey, we performed centration. Samples were size-fractionated by sieving and four experiments, in which we varied the respective mor- processed according to the ZooScan method (Gorsky et al., tality rate of large and small zooplankton (0.05 and 0.025 −3 −1 −1 2010). Observations included crustaceans, chaetognaths and [mmolNm ] d , respectively) by ±50 %: annelids greater than 500 µm. For model comparison we con- – A_high with 1.5 × µZ verted the observation from nighttime hauls to dry biomass L according to Lehette and Hernández-León(2009) and fur- × – A_low with 0.5 µZL ther to nitrogen units as suggested by Kiørboe(2013). A de- × × tailed description of the zooplankton processing is provided – B_high with 1.5 µZL and 1.5 µZS by Kiko and Hauss(2019). Only night observations were – B_low with 0.5 × µ and 0.5 × µ , compared since our model does not include diel vertical mi- ZL ZS gration. where µZi is the mortality rates of large and small zoo- In both the model and observations, concentration of large plankton. The average nitrogen flux to detritus due to large zooplankton is greatest in the surface and decreases with zooplankton mortality over the upper 100 m depth near the depth (Fig.2). At the surface, modelled concentrations are coast of Peru (coastal upwelling region; see Sect. 2.5 and 1 order of magnitude larger than observations at almost all −2 −1 · Fig.1) in the reference scenario is 3.1 mmolNm d (µZL stations (Fig.2). Only at station d do observations reach [Z ]2, Fig.3). Neglecting any non-linear and feedback ef- −3 L 1 mmolNm , while the model exhibits maximum values fects within the model, a 50 % change in the mortality rate −3 close to 4 mmolNm . At most stations, the distribution would result in a change in zooplankton loss due to mortality of modelled concentrations is similar to that of observed of 1.55 mmolNm−2 d−1. It is thus a conservative value, com- concentrations in the surface layer (upper 100 m), although pared to a change of 5 mmolNm−2 d−1 that anchovy popu- model estimates are consistently higher. This is also the case lation fluctuations could theoretically exert, as estimated in when comparing against a different dataset of observations Sect.1. (see Supplement). Below 100 m, however, model estimates All model experiments and the reference scenario were are consistently lower than the observations, which is in par- spun up for 30 climatological years. Annual means of the ticular evident at the deep offshore stations (AppendixB, state variables and nitrogen fluxes from the last climatologi- Fig. B1a and b). Zooplankton in our model does not con- cal year of the high-resolution domain were analysed. sume detritus or bacteria; small zooplankton feeds on phy- toplankton, and large zooplankton feeds on small zooplank- 2.5 Model analysis ton and on phytoplankton. Therefore, in contrast to observa- tions, its presence is not expected in deep water. In summary, The ETSP is highly dynamic at temporal and spatial scales, the model matches the observed spatial pattern of zooplank- with nutrient-rich cold water near the coast of Peru and olig- ton distribution but tends to overestimate zooplankton con- otrophic regions offshore. Therefore, we analysed three dif- centration in the surface layer and to underestimate it in the ferent regions: the “full domain” without boundaries (F), https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2896 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS

Figure 2. (a–e) Zooplankton concentrations (mmolNm−3). Lines indicate modelled large zooplankton concentrations in the reference sce- nario and experiments B_high and B_low averaged from January to March. Shaded area shows observed nighttime mesozooplankton biomass concentrations over the sampled depth intervals (m). Observations are lower than 0.1 mmolNm−3 below 200 m; thus they have not been in- cluded. For a plot including deep-water observations, please see AppendixB, Fig. B1, and Fig. 4 in Kiko and Hauss(2019). Bottom right: modelled large zooplankton biomass concentration at the surface in the reference scenario (mmolNm−3) and locations where observations were collected.

coastal upwelling area, where the concentration of large phy- toplankton is high, up to about 40 to 50 km offshore. Region O was picked to be as far as possible from the nutrient-rich areas along the Equator and along the coast but apart from the domain boundary to avoid boundary effects. For most of our analysis, percentage relative differences between the reference scenario and the other scenarios were calculated. In addition, we analyse the development of plankton succes- sion from the coast of Peru towards the open ocean at 12◦ S (Fig.1, white line). Because plankton concentrations are neg- ligible below 100 m depth and we are mainly interested in the plankton community, we focus most of our analysis on the Figure 3. Nitrogen flux from large (a) and small (b) zooplank- upper water layers. For our analysis we therefore integrate or ton to detritus due to zooplankton mortality, integrated over average plankton concentrations over the upper 100 m or, in the upper 100 m of the water column in the reference scenario the case of shallower waters, down to the seafloor. Also, all (mmolNm−2 d−1). analyses in our study take into account only annual averages. However, we recognise that there is high temporal variability in the NHCS (see AppendixC and Supplement). an “oligotrophic” region (O) offshore and the “coastal up- welling” region (C) near the Peruvian coast (Fig.1). Since the upwelling system of Peru is quite heterogeneous with lots of mesoscale processes, we restricted the C region to the very

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2897

3 Results

We first provide an overview of the general performance of the reference scenario, with respect to the different model components and biogeochemical provinces (Sect. 3.1). We then investigate their response to changes in zooplankton mortality and the response of the plankton ecosystem struc- ture (Sect. 3.2). The coastal upwelling region (C) is espe- cially productive and the habitat of the largest aggregations of small pelagic fish, whose temporal variability inspired this study. Therefore, in Sect. 3.3 we place special emphasis on this region. Finally, we investigate the response of the zoo- plankton losses due to mortality in the experiments, in or- der to understand whether the model structure buffers or in- creases the effect of varying the zooplankton mortality rate on such a term (Sect. 3.4). This would give us an insight into potential feedbacks to higher trophic levels.

3.1 Biogeochemistry and plankton distribution in the reference scenario

In this section we describe the concentrations of the inorganic and organic compartments in our model reference scenario, averaged between a 0 and 100 m depth (Fig.4) unless oth- Figure 4. Annually, spatially (oligotrophic (O) region, full domain erwise specified. Oxygen concentrations increase offshore, (F) and coastal upwelling (C) region; see Fig.1 for further refer- −3 with an average of 226.6 mmolO m−3 in the oligotrophic ence) and depth-averaged (0–100 m) concentrations (mmolNm 2 −3 −3 or mmolO2 m ) of the biogeochemical prognostic variables in region (O) compared to 69.7 mmolO2 m in the coastal upwelling region (Fig.4). The concentration decreases fur- the model reference scenario and normalised percent difference be- tween the reference scenario and experiments. P is phytoplankton; ther below 100 m, reaching an average of 5.3 mmolO m−3 2 Z is zooplankton; D is detritus; DON is dissolved inorganic nitro- between 100 and 1000 m in the coastal upwelling region gen; T is total; L is large; S is small. (C). Nitrate is the most abundant nutrient all over the do- main, ranging from 0.6 in O to 21.7 mmolNm−3 in C. On the other hand, ammonium and nitrite are only 0.8 and main (Fig.4), except for in the oligotrophic south-western − 3.2 mmolNm 3 in C, respectively (Fig.4). Please refer to region (Fig.4). Jose et al.(2017) for a further in-depth analysis of biogeo- The growth rate of phytoplankton is limited by tempera- chemical tracers in the reference scenario. ture, nutrients and light (see Eq.4). Hence, the spatial pattern Phyto- and zooplankton are generally absent in the deep of plankton near the coast can be explained by the compet- water. In the surface layer between a 0 and 100 m depth itive advantage of large phytoplankton in deep water due to (Fig.4), phytoplankton is clearly favoured by nutrient- its steeper initial slope of the P–I curve (see AppendixA, Ta- rich coastal upwelling, where total phytoplankton reaches ble A1), eutrophic conditions in the nutrient-rich upwelling − − 0.93 mmolNm 3 on average, compared to 0.25 mmolNm 3 water and relatively low predation due to the lack of large in the oligotrophic region (Fig.4). Total detritus follow the zooplankton. This opens a loophole for large phytoplankton concentration trend of plankton, with 0.26 in C and only 0.03 to grow in the upwelling waters. As water is transported off- in O. shore (Fig.1), large zooplankton starts to grow and grazes When zooming into the coastal region (Fig.5), large phy- on large phytoplankton. More oligotrophic sunlit conditions toplankton exhibits a sharp peak near the coast which drops even further offshore favour small phytoplankton growth at offshore. Moving further offshore, large zooplankton peaks the surface. Therefore, this first analysis reveals spatial seg- at the decline in the large phytoplankton peak, followed by regation and succession from the coast to offshore waters. increased concentrations of small phytoplankton. Given the These patterns are caused by the model’s parameterisation of (Ekman-driven) transport of surface waters (Fig.1) this spa- plankton groups and their mutual interactions. tial pattern might be interpreted as a form of succession as the water is advected offshore. In general, modelled concen- trations of large zooplankton are high not only in the coastal upwelling region (Sect. 2.3) but also in large parts of the do- https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2898 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS

Figure 5. Zonal distribution of surface plankton concentrations at 12◦ S annually averaged in the reference and the two B scenarios, respec- tively. The location of this section is indicated as a white line in Fig.1.

3.2 Response to zooplankton mortality inant group. However, its productivity increases by about 19 % in B_high and decreases by 22 % in B_low; i.e. its When changing zooplankton mortality, the inorganic vari- changes are much more pronounced than the overall phyto- ables (Fig.4) are not noticeably affected by the experiments. plankton response. Because small phytoplankton shows an Plankton responds very similarly to changes in mortality inverse response in production, this dampens the change in in experiments A and B (Fig.4 and AppendixD, Fig. D1). total primary production. Thus, a low zooplankton mortal- Phytoplankton and large zooplankton follow the same di- ity favours small phytoplankton and its growth, and a high rection of response in all regions: concentrations of large mortality favours large phytoplankton; changes in both phy- zooplankton decrease in the high mortality scenario and in- toplankton groups result in a weak response of total primary crease in the low mortality scenario, as could be expected. production. Large phytoplankton responds inversely to large zooplank- Experiment B_high exhibits the highest total plank- ton, evidencing a top-down control of its main grazer. On the ton biomass in the upper 100 m of the upwelling system other hand, the response of small phytoplankton is inverse to (112.6 mmolNm−2, Fig.6), which is mostly concentrated in that of large phytoplankton. In contrast, small zooplankton the large phytoplankton compartment (59.43 mmolNm−2). shows an asymmetric response to changes in mortality, as it In this experiment the main pathway of nitrogen transfer to mainly decreases in the low mortality scenarios but responds large zooplankton is via its grazing on large phytoplankton only weakly in the high mortality scenarios (Fig.4 and Ap- (6.77 mmolNm−2 d−1). As mortality decreases, small phy- pendixD, Fig. D1). toplankton and large zooplankton gain biomass. Large phyto- The spatial plankton distribution along the transect (Fig.5) plankton grazing remains the main nitrogen source for large remains the same when zooplankton mortality changes, but zooplankton. However, large zooplankton consumption of the absolute concentrations of each compartment change. In small phytoplankton is almost 3 times higher in B_low than all scenarios large phytoplankton peaks close to the coast. in B_high (Fig.6). Thus, a reduction in mortality causes a When large zooplankton concentrations are reduced because switch in the diet of large zooplankton, from mainly large of its higher mortality in experiment B_high, the large phyto- phytoplankton to a diet that consists of more than one-quarter plankton peak increases (Fig.5, right). Similarly, large phy- small phytoplankton. toplankton decreases with lower zooplankton mortality, due Small zooplankton biomass decreases by ∼ to higher grazing of zooplankton on phytoplankton (Fig.5, 0.4 mmolNm−2 (about 5 % of the reference value) in left). This pattern is also similar in experiments A. Because B_low, but it only increases by 0.15 mmolNm−2 (about the largest effects occur in the very productive coastal region 2 %) in B_high (Fig.6). Despite the changes in small (C), in the following section we narrow our analysis to this zooplankton biomass, the consumption of its biomass by domain. large zooplankton remains approximately the same in all experiments, resulting in a higher proportional biomass loss 3.3 Effects on the food web in the coastal domain of small zooplankton in scenario B_low. Hence, predation by large zooplankton as well as competition for food negatively An increase in zooplankton mortality causes only small affects small zooplankton. Under high mortality conditions, changes in total primary production in the coastal upwelling the availability of large phytoplankton as food increases region (C), but the partitioning between the two phytoplank- (Fig.6). However, small phytoplankton, the preferred prey ton groups changes (Fig.6). In particular, total primary pro- of small zooplankton, declines as explained above. Such ductivity of the system is increased by 3.9 % in B_high and reduced by 5.5 % in B_low. Large phytoplankton is the dom-

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2899

Figure 7. Percentage normalised difference in the nitrogen flux from large and small zooplankton to detritus due to zooplankton mortality, integrated over the upper 100 m of the water column, be- tween experiment B_high and experiment B_low and the reference scenario (see Fig.3 for the reference scenario).

antagonistic effects on small zooplankton buffer its response in this scenario.

3.4 Zooplankton losses due to mortality response

In this section, we describe the response of the nitrogen loss due to mortality, also referred to as the quadratic mortality ·[ ]2 ·[ ]2 term (µZi Zi ; see Eqs.5 and6). The integrated µZi Zi in the reference scenario was provided in Sect. 2.4, Fig.3. In ·[ ]2 ·[ ]2 experiment B, µZS ZS and µZL ZL exhibit different ·[ ]2 behaviours. In the coastal upwelling region, both µZS ZS ·[ ]2 and µZL ZL increase in B_high and decrease in B_low ·[ ]2 (Fig.6). However, the relative response of µZL ZL is mild and fluctuates between ±30 % outside the oligotrophic ·[ ]2 area (Fig.7). In contrast, µZS ZS exhibits a clear relative Figure 6. Concentrations (mmolNm−2) and nitrogen fluxes increase all over the domain in B_high, and a decrease in (mmolNm−2 d−1) between plankton compartments (small and B_low. The moderate response of large zooplankton loss can large phytoplankton PS and PL and small and large zooplankton be attributed to the combined effects of changes in zooplank- ZS and ZL, respectively) integrated over the upper 100 m or up to ton concentration ([ZL]) and changes in the mortality param- the seafloor if shallower than 100 m and averaged over latitude and ± eter (µZL ), which we varied by 50 % in this study. Large longitude in the coastal upwelling region (see Fig.1). SZ and SP in- zooplankton concentration increases when µZL is decreased, dicate the sinks which include phytoplankton mortality, zooplank- and vice versa. This opposite trend buffers the effect of a ton metabolism, large phytoplankton sedimentation, unassimilated change in µ . On the other hand, small zooplankton con- primary production and unassimilated grazing (Gutknecht et al., ZL centration ([Z ]) changes in the same direction as µ , due 2013a). S ZS to the combined effects of changes in its concentration, due to grazing pressure exerted by large zooplankton and compe-

tition for food with this group, and changes in µZS . https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2900 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS

To summarise, increasing and decreasing zooplankton the large zooplankton compartment of the ROMS–BioEBUS mortality by 50 % generates a rearrangement of the plankton ETSP configuration with mesozooplankton observations. ecosystem; however, the overall changes in the large zoo- At the surface, zooplankton concentrations simulated by plankton loss are not as high as would be expected from a our model in the reference scenario are 1 order of mag- change in the mortality rate alone. This might buffer the sys- nitude higher than observations at most stations. However, tem once the biogeochemical model is coupled to a model of sampling in the upper 10 m depth may be impacted by wa- higher trophic levels. ter disturbance by the ship adding additional errors to the measurements. The match to observations improves with depth. Modifying the mortality rate by +50 % (−50 %) pro- 4 Discussion duced only a change of −12 % (+19 %) in large zooplank- ton concentration, indicating that either the induced changes 4.1 Constraining the zooplankton compartment in mortality rate were not large enough or this parameter is not overly influential in improving the model fit to observa- An increasing need for the development of end-to-end mod- tions. Systems with a non-density-dependent, or linear, mor- els has generated interest in using results of biogeochemi- tality rate respond to perturbations in a “reactive” way, as cal models as forcing for higher-trophic-level models (fish, defined by Neubert et al.(2004), drifting away from equi- macroinvertebrates and apex predators) (see Tittensor et al., librium, in contrast to systems with density-dependent clo- 2018, for a review). In a one-way coupling set-up, the sure terms which tend to buffer the perturbations (Neubert biomass of plankton available as food for higher trophic lev- et al., 2004). Therefore, we might have expected a stronger els has been adjusted during calibration of the latter, reduc- impact if we had manipulated the linear closure term of the ing the amount of plankton that is available for fish con- model, or metabolic losses (see Sect. 2.2.2), rather than the sumption (e.g. Oliveros-Ramos et al., 2017; Travers-Trolet quadratic term. For an average nitrogen flux due to large zoo- et al., 2014b). However, for two-way coupling set-ups, this plankton mortality of 2.2, 3.1 and 3.52 mmolNm−2 d−1 and adjustment of the available plankton biomass could buffer large zooplankton integrated concentrations of 46.36, 38.82 the effect of higher trophic levels on lower trophic levels and 34.17 mmolNm−2 in the coastal upwelling region (see (e.g. Travers-Trolet et al., 2014b). Biogeochemical models Sect. 3.3), the mortality rate would be 0.04, 0.08 and 0.1 d−1 can produce a wide range of output depending on their pa- in scenarios B_low, reference and B_high, respectively. In rameter values (Baklouti et al., 2006) and their non-linearity. all cases the values are lower than the 0.19 d−1 estimated For example, a quadratic zooplankton mortality exacerbates by Hirst and Kiørboe(2002) for copepods in the field at the reduction in zooplankton biomass when concentrations 25 ◦C. The closest scenario to observations is B_high where are very high and prevents its extinction at very low con- the mortality rate is only about half of the estimate by Hirst centrations. In addition, the multiple-resource form of the and Kiørboe(2002). This is also the scenario that better re- Holling type-II grazing function allows the predator to mod- sembles mesozooplankton observations, since it exhibits the ify its grazing preference towards the most abundant prey lowest concentrations. On the other hand, the mortality rate (Fasham et al., 1999). Finally, Lima et al.(2002) noted that estimated for the reference scenario (0.08 d−1) is closer to the coupled physical and food web models can transition from 0.065 d−1 estimation by Hirst and Kiørboe(2002) at 5 ◦C. equilibrium to chaotic states under even small changes in This is considerably lower than the temperature in our mod- their parameters. Few studies have aimed to understand such elled region (see Fig.1). Note, however, that zooplankton behaviour (Baklouti et al., 2006) and examined the sensitiv- mortality in our model does not depend on temperature. ity of the model to parameters (Arhonditsis and Brett, 2004; Some part of the mismatch between model and observa- Shimoda and Arhonditsis, 2016). Our model study was partly tions might be related to how both data types are generated. motivated by the uncertainty associated with the zooplankton Therefore, a direct comparison between model and obser- mortality. Indeed our model showed that a small alteration in vations has to be viewed with some caution. In our model, the mortality parameter (small compared to the wide range large zooplankton acts as a closure term which is adjusted of values that have been used for this parameter in differ- to balance the biomass and nitrogen flux to other compart- ent biogeochemical models) can strongly affect the mass flux ments and does not resemble a specific set of species. Its pa- within the simulated ecosystem. Hence, there is an increas- rameters (maximum grazing rate, feeding preferences, etc.) ing need for accurate plankton representation in biogeochem- are meant to represent larger, slow-growing species with a ical models without dismissing other compartments, such as preference for diatoms. As such, they might not be directly nutrients or oxygen. Nevertheless, lack of data for valida- comparable to the observed groups. The observations, on the tion especially of higher trophic levels is a common problem other hand, are susceptible to sampling errors such as net for biogeochemical models of the northern Humboldt Up- avoidance and do not cover the whole taxonomic and size welling System (Chavez et al., 2008). Oxygen, chlorophyll spectrum of mesozooplankton. For instance, no gelatinous and nitrate in our model have been evaluated previously (Jose organisms are accounted for, which may account for an im- et al., 2017). Here we presented the first attempt to compare portant fraction of the wet biomass (Remsen et al., 2004);

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2901 only mesozooplankton greater than 500 µm is considered in To summarise, some biases and mismatches between the sampling (Kiko and Hauss, 2019); and fragile organ- model and observations remain; given the uncertainties and isms, such as Rhizaria, are not quantitatively sampled by nets episodic nature associated with the observations and their (Biard et al., 2016). Therefore, the observations might be bi- correspondence to their model counterparts, further studies ased low in comparison to the model. Furthermore, a lack will be necessary to more precisely calibrate the model. For of an explicit size term in the model limits a direct com- a complete model evaluation, however, the small zooplank- parison with observations because these depend on the mesh ton compartment should also be evaluated against microzoo- size of the sampling net. Finally, given the above-mentioned plankton samples. The high mortality scenario, B_high, is the rather pragmatic parameterisation especially of zooplankton one that is closest to the observations, due to producing the growth and loss rates, it is very likely that the model could smallest concentrations of large zooplankton at the surface. be improved by a tuning or calibration exercise that targets However, changing this parameter was obviously not enough a good match between observed and simulated zooplankton to match the observations. In fact, in our model an increase concentrations. Both the mortality estimation by Hirst and of 50 % in the mortality rate produced only an increase of Kiørboe(2002) and the zooplankton observations in the field 14 % (0.4 mmolNm−2 d−1) in large zooplankton mortality suggest that further tuning of the model should lean towards loss (see Sect.3) because of the high non-linearity of the higher mortality rates. Nevertheless, this may require the fur- model. Indeed, potential changes in zooplankton losses of ther tuning of other parameters. Despite the complexity of 5 mmolNm−2 d−1, derived by fluctuations in anchovy stocks the model, the considerable uncertainty in model parameters and grazing pressure (see Sect.1), point towards much larger and the sparsity of observations that can constrain these pa- values for the mortality rate. An even stronger increase in the rameters, this is a complex task (see e.g. Kriest et al., 2017). zooplankton mortality rate (e.g. Lima and Doney, 2004, ap- Therefore, we have refrained from this effort for the present plied a 5-times-larger value), along with a subsequent adjust- but aim at providing a better-calibrated model in the future. ment of other parameters, may be necessary to approach ob- The spatial variability between different profiles of zoo- served values. In addition, complementary observations with plankton is greater in the observations than in the model, other sampling methods could provide a better estimation of and the variability in concentrations within each single pro- mesozooplankton concentrations for tuning the model. file is much larger than the differences between the modelled mortality scenarios (Fig.2). Several sources of variability are 4.2 Zooplankton mortality and the response of the not accounted for in the model as it only simulates the most pelagic ecosystem relevant processes in the system. We employ a climatolog- ical model which aims at simulating the average dynamics over several years, dismissing interannual variability. Fur- Our model study showed the strongest response of the thermore, we here compare a 3-month average from January ecosystem to changes in the mortality rate in the highly pro- to March, while observations provide only a snapshot of a ductive coastal upwelling. Here, the response of the model highly dynamical system. In addition, we only compared our ecosystem was mainly driven by large zooplankton. This simulated zooplankton against night observations because in can be concluded from the close similarity of model solu- our model zooplankton is always active at the surface. In real- tions A_high and B_high, as well as of A_low and B_low ity, zooplankton is known to perform diel vertical migrations (see AppendixD). The mortality term for small zooplank- (DVMs), which could increase the export flux to the deep ton played a lesser role; in addition to the direct effect of the ocean (Aumont et al., 2018; Archibald et al., 2019; Kiko and mortality rate, this compartment was also affected by graz- Hauss, 2019; Kiko et al., 2020). The lack of DVM could af- ing by, and competition with, large zooplankton. Large zoo- fect the export of organic matter to greater depths and there- plankton fluctuations due to mortality directly affected large fore the biogeochemical turnover at the surface. Zooplank- phytoplankton through grazing. Small phytoplankton, on the ton likely also experiences lower mortality at depth (Ohman, other hand, was affected by grazing but also by competi- 1990); however, off Peru these benefits might be counter- tion with large phytoplankton. Changing the mortality rate balanced by reduced oxygen availability and the concurrent produced little effect on the mass loss of large zooplankton metabolic costs. These obstacles for comparing zooplank- due to mortality; however, it altered the nitrogen pathways ton models with observations had already been described by along the trophic chain and ultimately the concentrations of Mullin(1975) more than 4 decades ago: (a) “the zooplank- most plankton groups, albeit in different ways, depending ton is a very heterogeneous group, defined operationally by on the direction of change. Under conditions of high zoo- the gear used for capture rather than by a discrete position in plankton mortality the food chain is dominated by nitrogen the food web” (Mullin, 1975). (b) Zooplankton is irregularly transfer from large phytoplankton to large zooplankton, the distributed in space, not necessarily following physical fea- classical food web attributed to highly productive upwelling tures. (c) Adult stages of some zooplankton groups perform systems (Ryther, 1969). When zooplankton mortality is re- vertical migrations (Mullin, 1975). duced, large zooplankton increases its consumption of small phytoplankton, taking over the role of small zooplankton. https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2902 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS

In our model, large zooplankton has a competitive ad- Table 1. Qualitative comparison of the response of total, large and vantage by feeding on its competitor, small zooplankton, a small zooplankton (ZT, ZL, ZS) and phytoplankton (PT, PL, PS) strategy that was also found to evolve in simple ecosystem to a 50 % higher and lower zooplankton mortality parameter in our models as an advantageous alternative to direct competition experiments B, with the results from Getzlaff and Oschlies(2017) (Cropp and Norbury, 2020). We find that under low mortal- (ZGO, PGO). Full, oligotrophic and coastal upwelling refer to the ity conditions, this advantage increases. The importance of regions in our study (see Fig.1) integrated over the upper 100 m; global, Southern Ocean and tropics refer to the study by Getzlaff competition is further evidenced in the changes in small phy- and Oschlies(2017, their Figs. 2 and 3 at year 300). We grouped to- toplankton concentrations in the coastal upwelling region. gether global and full because both refer to the whole model domain These were partly driven by changes in the availability of in each of the two studies. Similarly, oligotrophic and tropics refer resources arising from fluctuations in large phytoplankton to regions characterised by low nutrient concentrations, and coastal concentrations, which constitute the dominant group in the upwelling and Southern Ocean are both regions with high nutrient coastal upwelling. Natural selection, competitive exclusion concentrations. and different resource utilisation strategies, together with bottom-up forcing by the physical processes in the environ- ZT PT ZL PL ZS PS ZGO PGO ment, can shape the plankton community in global models Full and global (Follows et al., 2007; Dutkiewicz et al., 2009; Barton et al., 2010) and indicate bottom-up effects on the phytoplankton High − + − + − − − + + − + − − + + − community. On the other hand, Prowe et al.(2012) showed Low that variable zooplankton predation can increase phytoplank- Oligotrophic and tropics ton diversity by opening refuges for less competitive phy- High − + − + − + − − toplankton groups and thus exert top-down effects. In our Low + − + − + + + + study, the biological interactions between two phytoplank- ton groups, mainly competition for resources (bottom-up), Coastal upwelling and Southern Ocean are additionally affected in a top-down manner by changes High − + − + + − + + in zooplankton concentrations. Low + − + − − + − − The processes driving the ecosystem response in our re- gional study are dominated by trophic interactions among the size classes of phytoplankton and zooplankton. We found a top-down-driven response affecting mainly the plank- both studies, but they are slightly milder in the study by Get- ton compartments of the model. The direction of the total zlaff and Oschlies(2017). At the regional scale, the responses zooplankton and total phytoplankton change is determined in the study of Getzlaff and Oschlies(2017) have a feed- by the large zooplankton and large phytoplankton groups. back from lower trophic levels, either to phytoplankton in the Small zoo- and phytoplankton buffer the response when they nutrient-replete region or all the way to nutrients in the olig- present opposite trends to their larger counterparts (Table1). otrophic region. Therefore, the biomass of both zooplank- In the following, we will compare our results with a similar ton and phytoplankton is ultimately bottom-up affected. This study by Getzlaff and Oschlies(2017). is evidenced by a change in zooplankton and phytoplankton Getzlaff and Oschlies(2017), in their sensitivity study, concentrations in the same direction (Getzlaff and Oschlies, modified the zooplankton mortality by ±50 % in a global 2017, their Fig. 3). On the other hand, although our study biogeochemical model with a spin-up time of 300 years. also exhibits a feedback from phytoplankton to zooplankton, Their model has only one zooplankton size class with a the strongest driver remains the top-down predation of zoo- quadratic mortality term. For their analysis, Getzlaff and Os- plankton on phytoplankton (Table1). chlies(2017) evaluated three regions: the whole domain, The regional differences between our study and that by also referred to here as “global”; the region from 20◦ S to Getzlaff and Oschlies(2017) can likely be explained by the 20◦ N which in this coarse-resolution model is mainly an different model structures and experimental set-ups, namely oligotrophic region and is referred to as “tropics”; and the the number of phyto- and zooplankton compartments, differ- region south of 40◦ S where nutrient concentrations are high ent timescales considered for model simulation and analysis, and is named “Southern Ocean”. After the model spin-up, and the spatial domain: while Getzlaff and Oschlies(2017) global zooplankton biomass changes by about −(+)5 %, and applied a global biogeochemical model with just one size phytoplankton biomass by about +(−)1 % in the high (low) class of phyto- and zooplankton, simulated until near a steady mortality scenario (Getzlaff and Oschlies, 2017, their Fig. 2). state, our regional model study applies a more complex bio- In contrast, in our study, total zooplankton averaged over the geochemical model with a strong seasonal cycle (see Ap- full domain decreases (increases) by −11 % (10 %), and total pendixC), simulated for only 30 years and constantly forced phytoplankton changes by about +(−)6 % in the high (low) at the boundaries. Further, the short few-year timescale of mortality scenario (Fig.4 and Table1). Hence, the effects our model simulations might have prevented the effects of of changing zooplankton mortality follow the same trend in changed zooplankton mortality from propagating to deeper

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2903 water layers which contain the largest concentrations of nu- plankton loss due to mortality between the low and refer- trients. Finally, the region modelled in our study is spatially ence scenario and between the high and reference scenario, already very dynamic at the mesoscale resolution, as evi- respectively. As shown in Sect. 3.4, the mortality rate in our denced by well-defined plankton spatial succession from the study is also smaller than the ±50 % changes that would be coast of the continent towards the open ocean. The phyto- expected from a change in the mortality parameter of ±50 % plankton bloom, which develops closest to the coast and then (see Sect. 2.4). The non-linearity of both global and regional is offset while water is transported offshore, can be explained models seems to reduce the effect of changes in the mortality by an imbalance between sources and sinks, triggered by parameter on zooplankton loss. changing environmental conditions. For example, Irigoien In summary both studies show a similar global response et al.(2005) applied the concept of “loopholes”, proposed by to changes in zooplankton mortality, driven by zooplankton Bakun and Broad(2003) to explain fish productivity and re- preying on phytoplankton. Two zooplankton and phytoplank- cruitment success, to phytoplankton: according to their con- ton size classes present opposite trends in our studies, buffer- cept, phytoplankton blooms are formed when environmental ing the overall response. Nevertheless, the relative changes conditions open a loophole in the plankton community of a in total zooplankton and total phytoplankton are on the same mature ecosystem. Then, particularly phytoplankton species order of magnitude as in Getzlaff and Oschlies(2017) and that are able to escape microzooplankton predation are those even slightly higher. Regionally different feedbacks operated that will take advantage of the loophole and bloom (Irigoien in the two models, possibly due to the specific set-up of et al., 2005). Our model results suggest that similar processes each study, spin-up time and resolution. Finally, the relative occur. Low concentrations of large zooplankton allow large change in the zooplankton loss due to mortality is smaller phytoplankton to bloom near the coast. While the water is than the expected ±50 % in both studies. advected offshore, zooplankton growth and grazing offset the bloom. Observations by Franz et al.(2012) also reported spa- 4.3 From plankton to higher trophic levels tial succession with large diatoms abundant in the coastal upwelling region being replenished by nanophytoplankton In our study, we changed the zooplankton quadratic mortal- offshore. However, they propose silicate as the limiting nu- ity, which could be regarded as the effect of a predator tar- trient offsetting the diatoms’ peak, which is not present in geting highly aggregated zooplankton populations, or whose our model. Furthermore, such succession is not unique to concentration closely follows that of zooplankton. This can the NHCS but a characteristic feature of upwelling regions be viewed as a way to parameterise the effect of changing (Hutchings, 1992). We note that in the present configura- fish abundance on the biogeochemistry of the system. In tion the BioEBUS model does not include any temperature- this case, a low zooplankton mortality implies fewer small dependent zooplankton grazing rate. We expect that, if this pelagic fish (such as anchovies and sardines), while a high were implemented, the loophole for phytoplankton growth zooplankton mortality implies a higher abundance of such in the cold waters of the coastal upwelling region would fish. Further, our experiments are based on two different as- even be widened, amplifying the spatial succession we ob- sumptions: one where small pelagic fish feed only on large served. However, temperature might also affect zooplank- zooplankton (experiments A) and one where they feed on ton metabolism, with colder temperatures decreasing its loss and affect the mortality of both large and small zooplankton rates, which could in turn mute these effects again. On the (experiments B). other hand, the global model by Getzlaff and Oschlies(2017) The diet of anchovy is still under debate. While previ- does not resolve mesoscale processes. Furthermore, while ous studies had considered that anchovies feed mainly on they divided their study into the tropics, as an oligotrophic phytoplankton, Espinoza and Bertrand(2008) concluded that region, and the Southern Ocean, as an upwelling region, the anchovies feed on zooplankton, especially euphausiids and upwelling system off Peru in the eastern tropical South Pa- copepods. Furthermore, the diet of anchovy seems to be more cific is a nutrient-rich area. For all of this, we based our com- flexible than previously considered (Espinoza and Bertrand, parison on similarities in the nutrient concentration (high nu- 2008). For instance, the anchovy collapse in 1972 was corre- trients, oligotrophic and whole domain), rather than on geo- lated with a shift from a population feeding mostly on phy- graphic overlap. toplankton to a southern population feeding on zooplank- In Getzlaff and Oschlies(2017), the zooplankton loss ton (Hutchings, 1992). On the other hand, small zooplankton due to mortality is not provided. However, it can be cal- groups such as ciliates have been reported as a minor com- culated from the zooplankton concentration and mortality ponent of anchovy diet (Table 5 in Espinoza and Bertrand, rate. Assuming integrated zooplankton biomasses at year 300 2008). Thus, experiments A are more likely to resemble the (Getzlaff and Oschlies, 2017, their Fig. 2) of 98, 93 and fluctuations in anchovy populations. On the other hand, sar- 89 Tg N in the low mortality, reference and high mortality dines, with their finer gill rakers, obtain most of their nu- scenarios, respectively; mortality parameters of 0.03, 0.06 trition from microzooplankton (van der Lingen et al., 2006). and 0.09 (mmolNm3)−1 d−1; and a quadratic mortality term, Although currently sardines are not as abundant as anchovies then there is a difference of −44.5 % and 37.4 % in the zoo- off Peru, historically they have also built up large concen- https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2904 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS trations and strong fluctuations over time have been ob- 5 Conclusions served (Lluch-Belda et al., 1989; Rykaczewski and Check- ley, 2008). Thus, when also considering sardine populations In summary, our study showed that changes in zooplankton and feeding modes, experiments B (simultaneous mortality mortality can have a strong impact on the trophic interac- change in both large and small zooplankton) might be more tions between the plankton compartments of the model. Such appropriate to parameterise the effects of changed fishing changes are meant to mimic variations in the abundance of mortality on lower trophic levels. planktivorous fish in the system. Large zooplankton mortal- Although the quadratic mortality rate is constant over the ity, as the top predator in the model, is the main driver of entire domain, the fractional loss rate by zooplankton mortal- the planktonic food web response in the model. Changes in × −1 the mortality rate of small zooplankton, which may resem- ity (µZi Zi; d ) varies over the domain because of changes in zooplankton concentration. This might mimic spatially ble fluctuations in the sardine populations, are masked when variable grazing pressure by fish. However, our experimen- large zooplankton mortality also changes. The zooplankton tal set-up might be too simple to investigate the detailed high mortality scenario, which mimics an increase in plank- response of predator–prey relationships. We partly tried to tivorous fish, generates a shorter food web where most of avoid conclusions that are too general by focusing our analy- the nutrients are taken up by large phytoplankton. In the low sis on the coastal upwelling region off Peru, since anchovies mortality scenario, the biomass of small phytoplankton in- are highly concentrated in this region (Checkley et al., 2009). creases and a longer food chain where nitrogen reaches large In addition, we neglected any feedback effects between zoo- zooplankton through consumption of small zooplankton is plankton and their predator fish. A more detailed model set- favoured (Sect. 3.3). Our 50 % mortality changes are small up, as, for example, in coupled biogeochemical–fish models compared to changes expected from the population fluctua- (Oliveros-Ramos et al., 2017), would help to elucidate the tions that small pelagic fish have historically experienced in specific trophic interactions in this region and their response the NHCS (Sects.1 and 2.4). In this study we were inter- to environmental changes and changes in fishing pressure. ested in the possible top-down effects of those fluctuations Finally, we note that the parameterisation of fish preda- on the biogeochemistry, rather than on their causes. How- tion in our study implies that all organic matter is transferred ever, in a highly bottom-up-driven system, it is important to from zooplankton straight to the detritus compartment and be cautious and conservative when evaluating top-down sce- then remineralised. In the real world, it is stored as fish for narios. A fully coupled end-to-end ecosystem model explic- some time until the fish defecate or die (Allgeier et al., 2017). itly including fish (as by, for example, Travers-Trolet et al., In an equilibrium state this would not be a problem since 2014b) would allow the representation of the effect of tem- there is a constant turnover of nutrients. In a non-equilibrium poral and spatial variability of fish. It would also allow for state, the time that nutrients spend as part of larger ani- a specialised targeting of fish food and for including the mals’ biomass would increase the gap between nutrient con- bottom-up effect of changing zooplankton concentration on sumption by phytoplankton and their replenishment, affect- fish populations, as well as their top-down effect and its po- ing phytoplankton growth rate and potentially the blooming tential consequences for the entire ecosystem. However, this timing. However, this should not be a problem in the coastal would also involve the inclusion of more parameters in the upwelling region because nutrients are highly concentrated model (up to hundreds of parameters; see Oliveros-Ramos here. This is not the case for the oligotrophic region although, et al., 2017), which are only poorly constrained. Therefore, in our study, this region presents the weakest response. Fur- while explicitly including fish in a model widens the possi- thermore, small pelagic fish concentrate mainly in the highly bilities for controlling the system, it may also increase the productive upwelling region rather than in the oligotrophic sources of uncertainty. Here we utilised an already-validated waters offshore. On the other hand, fish and larger animals physical and biogeochemical model and parameterised the are highly mobile and do not constantly drift by advection loss of zooplankton due to fluctuations in small pelagic fish, as nutrients and plankton do. Therefore, migrations transport without adding additional complexity to the model. Our re- nutrients and organic matter in and out of the region in a sults may be a baseline reference for further studies exploring horizontal (McInturf et al., 2019; Varpe and Fiksen, 2005; such an effect. Williams et al., 2018) and vertical (Davison et al., 2013; Lav- ery et al., 2010) fashion. Such nutrient dynamics over time and space driven by higher trophic levels can be further ex- plored either in explicitly coupled end-to-end models or in an implicit way similarly to in our study.

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2905

Appendix A: Plankton parameters

Our reference simulation has the same configuration as in Jose et al.(2017) which in turn is based on the parameters used in Gutknecht et al.(2013a), with minor adjustments. Ta- ble A1 provides a list of the most relevant parameters for this study, as well as their descriptions and units. This is not a comprehensive list; for the full list of parameters and their values please see Gutknecht et al.(2013a) and Jose et al. (2017).

https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2906 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS

Table A1. Plankton parameters in the model. The complete list of biogeochemical parameters is given in Gutknecht et al.(2013a), and the values for the reference simulation are provided in Jose et al.(2017).

Symbol Description Value Unit −2 −1 −1 αPS Initial slope of P–I curve for PS 0.025 (Wm ) d −2 −1 −1 αPL Initial slope of P–I curve for PL 0.04 (Wm ) d −1 µPS Mortality rate of PS 0.027 d −1 µPL Mortality rate of PL 0.03 d −3 KNH Half-saturation constant for uptake of NH by P 0.5 mmolNm 4PS 4 S K Half-saturation constant for uptake of NH by P 0.7 mmolNm−3 NH4PL 4 L −3 KNO Half-saturation constant for uptake of NO3 + NO2 by PS 1.0 mmolNm 3PS −3 K Half-saturation constant for uptake of NO + NO by PL 2 mmolNm NO3PL 3 2 −1 wPL Sedimentation velocity of PL 0.5 md f 1ZS Assimilation efficiency of ZS 0.75 – f 1ZL Assimilation efficiency of ZL 0.7 – g Maximum grazing rate of Z 0.9 d−1 maxZS S g Maximum grazing rate of Z 1.2 d−1 maxZL L eZSPS Preference of ZS for PS 0.75 – eZSPL Preference of ZS for PL 0.25 – eZLPS Preference of ZL for PS 0.26 – eZLPL Preference of ZL for PL 0.5 – eZLZS Preference of ZL for ZS 0.24 – −3 kZS Half-saturation constant for ingestion by ZS 1 mmolNm −3 kZL Half-saturation constant for ingestion by ZL 2 mmolNm −3 −1 −1 µZS Mortality rate of ZS 0.025 (mmolNm ) d −3 −1 −1 µZL Mortality rate of ZL 0.05 (mmolNm ) d −1 γZS Metabolic rate of ZS 0.05 d −1 γZL Metabolic rate of ZL 0.05 d

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2907

Appendix B: Mesozooplankton evaluation

Modelled large zooplankton is higher than observed night- time mesozooplankton above 100 m (Fig. B1). Deep-water zooplankton is absent in the model, while observed mesozoo- plankton is present at depths between 600 and 1000 m. Meso- zooplankton below 200 m further increases during the day- time (data not shown; please see Fig. 4 in Kiko and Hauss, 2019).

https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2908 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS

Figure B1. Same as Fig.2, but (a–e) show a depth range up to 1000 m and a logarithmic horizontal axis.

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2909

Appendix C: Temporal variability

The NHCS presents high climatological and interannual vari- ability. In our modelled region, small phytoplankton concen- trations are relatively stable throughout the year. On the other hand, large phytoplankton production exhibits a clear sea- sonal pattern, with the largest concentrations presented in austral summer (Fig. C1). Echevin et al.(2008) discuss a sim- ilar seasonal pattern found in their model. Such a pattern has also been identified in satellite-derived primary production (Messié and Chavez, 2015).

https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2910 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS

Figure C1. Primary production by large and small phytoplankton during the last climatological year of the simulation, averaged over the upwelling region (see Fig.1) and integrated over the upper 100 m in the reference scenario.

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2911

Appendix D: Plankton surface concentrations

Experiments A and B exhibit very similar spatial trends. Figure D1 provides an overview of surface plankton con- centrations in their reference scenario and their changes in the experiments. Note that in the two sets of experiments, large zooplankton increased when mortality decreased and vice versa. On the other hand, small zooplankton presents a counter-intuitive response when mortality decreases, re- sponding to the change in the concentration of its preda- tor rather than to changes in mortality, and an ambiguous response in the high mortality cases. Large and small phy- toplankton exhibit opposite trends, seemingly driven by the concentration change of their main predator (large and small zooplankton, respectively).

https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2912 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS

Figure D1. Large and small zooplankton (ZS and ZL) and large and small phytoplankton (PL and PS) integrated over the upper 100 m of the water column (mmolNm−2). Rows from top to bottom: reference scenario and difference between experiments A_high, B_high, A_low and B_low and the reference scenario.

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2913

Code and data availability. Code and data used in this study are Anderson, T. R., Gentleman, W. C., and Sinha, B.: Influ- available upon request. The ROMS model is maintained at https: ence of grazing formulations on the emergent properties of //www.myroms.org (last access: 22 April 2021). a complex ecosystem model in a global ocean general cir- culation model, 3rd GLOBEC OSM: From ecosystem func- tion to ecosystem prediction, Prog. Oceanogr., 87, 201–213, Supplement. The supplement related to this article is available on- https://doi.org/10.1016/j.pocean.2010.06.003, 2010. line at: https://doi.org/10.5194/bg-18-2891-2021-supplement. Archibald, K. M., Siegel, D. A., and Doney, S. C.: Modeling the Impact of Zooplankton Diel Vertical Migration on the Carbon Export Flux of the Biological Pump, Global Biogeochem. Cy., Author contributions. IK and AO designed the study. YSJ carried 33, 181–199, https://doi.org/10.1029/2018GB005983, 2019. out the simulations. MHC performed the analysis. RK and HH pro- Arhonditsis, G. B. and Brett, M. T.: Evaluation of the current state of vided zooplankton observations and expertise on zooplankton dy- mechanistic aquatic biogeochemical modeling, Mar. Ecol. Prog. namics in the NHCS. All authors discussed the results and wrote Ser., 271, 13–26, https://doi.org/10.3354/meps271013, 2004. the manuscript. Aumont, O., Ethé, C., Tagliabue, A., Bopp, L., and Gehlen, M.: PISCES-v2: an ocean biogeochemical model for carbon and ecosystem studies, Geosci. Model Dev., 8, 2465–2513, https://doi.org/10.5194/gmd-8-2465-2015, 2015. Competing interests. The authors declare that they have no conflict Aumont, O., Maury, O., Lefort, S., and Bopp, L.: Evaluating the Po- of interest. tential Impacts of the Diurnal Vertical Migration by Marine Or- ganisms on Marine Biogeochemistry, Global Biogeochem. Cy., 32, 1622–1643, https://doi.org/10.1029/2018GB005886, 2018. Acknowledgements. We would like to thank Tianfei Xue for car- Ayón, P., Criales-Hernandez, M. I., Schwamborn, R., and Hirche, rying out some of the model simulations. We also thank the three H.-J.: Zooplankton research off Peru: A review, Prog. Oceanogr., anonymous reviewers for their very helpful and constructive com- 79, 238–255, https://doi.org/10.1016/j.pocean.2008.10.020, ments. Simulations were performed using the computing facilities 2008a. of the Norddeutscher Verbund zur Förderung des Hoch- und Höch- Ayón, P., Swartzman, G., Bertrand, A., Gutiérrez, M., and Bertrand, stleistungsrechnens – HLRN. S.: Zooplankton and forage fish species off Peru: large-scale bottom-up forcing and local-scale depletion, Prog. Oceanogr., 79, 208–214, https://doi.org/10.1016/j.pocean.2008.10.023, Financial support. This research has been supported by the 2008b. Bundesministerium für Bildung und Forschung projects Biard, T., Stemmann, L., Picheral, M., Mayot, N., Vandromme, P., CUSCO and Humboldt Tipping (grant no. 03F0813A and Hauss, H., Gorsky, G., Guidi, L., Kiko, R., and Not, F: In situ grant no. 01LC1823B). Rainer Kiko was further supported via a imaging reveals the biomass of giant protists in the global ocean, Make Our Planet Great Again grant of the French National Re- Nature, 532, 504–507, https://doi.org/10.1038/nature17652, search Agency within the Programme d’Investissements d’Avenir, 2016. reference ANR-19-MPGA-0012. Further support for this project Baklouti, M., Faure, V., Pawlowski, L., and Sciandra, A.: Investi- was provided by the Deutsche Forschungsgemeinschaft (DFG), gation and sensitivity analysis of a mechanistic phytoplankton under the Sonderforschungsbereich 754 “Climate-Biogeochemistry model implemented in a new modular numerical tool (Eco3M) Interactions in the Tropical Ocean” project (http://www.sfb754.de, dedicated to biogeochemical modelling, Prog. Oceanogr., 71, last access: 3 May 2021). 34–58, https://doi.org/10.1016/j.pocean.2006.05.003, 2006. Bakun, A. and Broad, K.: Environmental ‘loopholes’ and fish pop- The article processing charges for this open-access ulation dynamics: comparative pattern recognition with focus on publication were covered by a Research El Niño effects in the Pacific, Fish. Oceanogr., 12, 458–473, Centre of the Helmholtz Association. https://doi.org/10.1046/j.1365-2419.2003.00258.x, 2003. Barton, A. D., Dutkiewicz, S., Flierl, G., Bragg, J., and Follows, M. J.: Patterns of Diversity in Marine Phytoplankton, Science, Review statement. This paper was edited by Stefano Ciavatta and 327, 1509–1511, https://doi.org/10.1126/science.1184961, 2010. reviewed by three anonymous referees. Beddington, J. R. and May, R. M.: Harvesting Natural Populations in a Randomly Fluctuating Environment, Science, 197, 463–465, https://doi.org/10.1126/science.197.4302.463, 1977. Carton, J. A. and Giese, B. S.: A reanalysis of ocean climate using References Simple Ocean Data Assimilation (SODA), Mon. Weather Rev., 136, 2999–3017, https://doi.org/10.1175/2007MWR1978.1, Alheit, J. and Niquen, M.: Regime shifts in the Hum- 2008. boldt Current ecosystem, Prog. Oceanogr., 60, 201–222, Chavez, F. P. and Messié, M.: A comparison of Eastern https://doi.org/10.1016/j.pocean.2004.02.006, 2004. Boundary Upwelling Ecosystems, Prog. Oceanogr., 83, 80–96, Allgeier, J. E., Burkepile, D. E., and Layman, C. A.: Animal https://doi.org/10.1016/j.pocean.2009.07.032, 2009. pee in the sea: consumer-mediated nutrient dynamics in the Chavez, F. P., Ryan, J., Lluch-Cota, S. E., and Ñiquen, M.: world’s changing oceans, Glob. Change Biol., 23, 2166–2178, From Anchovies to Sardines and Back: Multidecadal https://doi.org/10.1111/gcb.13625, 2017. https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2914 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS

Change in the Pacific Ocean, Science, 299, 217–221, the oceanic mixed layer, J. Mar. Res., 48, 591–639, https://doi.org/10.1126/science.1075880, 2003. https://doi.org/10.1357/002224090784984678, 1990. Chavez, F. P., Bertrand, A., Guevara-Carrasco, R., Soler, P., and Fasham, M. J. R., Boyd, P. W., and Savidge, G.: Modeling the Csirke, J.: The northern Humboldt Current System: Brief history, relative contributions of autotrophs and heterotrophs to car- present status and a view towards the future, Prog. Oceanogr., 79, bon flow at a Lagrangian JGOFS station in the Northeast At- 95–105, https://doi.org/10.1016/j.pocean.2008.10.012, 2008. lantic: the importance of DOC, Limnol. Oceanogr., 44, 80–94, Checkley, D. M., Ayon, P., Baumgartner, T. R., Bernal, M., https://doi.org/10.4319/lo.1999.44.1.0080, 1999. Coetzee, J. C., Emmett, R., Guevara-Carrasco, R., Hutch- Fennel, K., Wilkin, J., Levin, J., Moisan, J., O’Reilly, J., and Haid- ings, L., Ibaibarriaga, L., Nakata, H., Oozeki, Y., Planque, vogel, D.: Nitrogen cycling in the Middle Atlantic Bight: Results B., Schweigert, J., Stratoudakis, Y., and van der Lingen, from a three-dimensional model and implications for the North C. D.: Habitats, in: Climate Change and Small Pelagic Fish, Atlantic nitrogen budget, Global Biogeochem. Cy., 20, GB3007, edited by Checkley, D. M., Alheit, J., Oozeki, Y., and Roy, https://doi.org/10.1029/2005GB002456, 2006. C., chap. 3, 64–146, Cambridge University Press, Cambridge, Follows, M. J., Dutkiewicz, S., Grant, S., and Chisholm, https://doi.org/10.1017/CBO9780511596681, 2009. S. W.: Emergent Biogeography of Microbial Commu- Cropp, R. and Norbury, J.: The emergence of new trophic levels in nities in a Model Ocean, Science, 315, 1843–1846, eco-evolutionary models with naturally-bounded traits, J. Theor. https://doi.org/10.1126/science.1138544, 2007. Biol., 496, 110 264, https://doi.org/10.1016/j.jtbi.2020.110264, Franz, J., Krahmann, G., Lavik, G., Grasse, P., Dittmar, T., and 2020. Riebesell, U.: Dynamics and stoichiometry of nutrients and phy- Cury, P., Bakun, A., Crawford, R. J., Jarre, A., Quinones, R. A., toplankton in waters influenced by the oxygen minimum zone Shannon, L. J., and Verheye, H. M.: Small pelagics in up- in the eastern tropical Pacific, Deep-Sea Res. Pt. I, 62, 20–31, welling systems: patterns of interaction and structural changes https://doi.org/10.1016/j.dsr.2011.12.004, 2012. in “wasp-waist” ecosystems, ICES J. Mar. Sci., 57, 603–618, Getzlaff, J. and Oschlies, A.: Pilot Study on Potential Impacts of https://doi.org/10.1006/jmsc.2000.0712, 2000. Fisheries-Induced Changes in Zooplankton Mortality on Ma- Daewel, U., Hjøllo, S. S., Huret, M., Ji, R., Maar, M., Niira- rine Biogeochemistry, Global Biogeochem. Cy., 31, 1656–1673, nen, S., Travers-Trolet, M., Peck, M. A., and van de Wolf- https://doi.org/10.1002/2017GB005721, 2017. shaar, K. E.: Predation control of zooplankton dynamics: a re- Gorsky, G., Ohman, M. D., Picheral, M., Gasparini, S., Stemmann, view of observations and models, ICES J. Mar. Sci., 71, 254–271, L., Romagnan, J.-B., Cawood, A., Pesant, S., García-Comas, https://doi.org/10.1093/icesjms/fst125, 2014. C., and Prejger, F.: Digital zooplankton image analysis using Davison, P. C., Checkley, D. M., Koslow, J. A., and Bar- the ZooScan integrated system, J. Plankton Res., 32, 285–303, low, J.: Carbon export mediated by mesopelagic fishes in https://doi.org/10.1093/plankt/fbp124, 2010. the northeast Pacific Ocean, Prog. Oceanogr., 116, 14–30, Gutknecht, E., Dadou, I., Le Vu, B., Cambon, G., Sudre, J., Garçon, https://doi.org/10.1016/j.pocean.2013.05.013, 2013. V., Machu, E., Rixen, T., Kock, A., Flohr, A., Paulmier, A., and Dutkiewicz, S., Follows, M. J., and Bragg, J. G.: Mod- Lavik, G.: Coupled physical/biogeochemical modeling includ- eling the coupling of ocean ecology and biogeo- ing O2−dependent processes in the Eastern Boundary Upwelling chemistry, Global Biogeochem. Cy., 23, GB4017, Systems: application in the Benguela, Biogeosciences, 10, 3559– https://doi.org/10.1029/2008GB003405, 2009. 3591, https://doi.org/10.5194/bg-10-3559-2013, 2013a. Echevin, V., Aumont, O., Ledesma, J., and Flores, G.: The sea- Gutknecht, E., Dadou, I., Marchesiello, P., Cambon, G., Le Vu, B., sonal cycle of surface chlorophyll in the Peruvian upwelling Sudre, J., Garçon, V., Machu, E., Rixen, T., Kock, A., Flohr, system: A modelling study, Prog. Oceanogr., 79, 167–176, A., Paulmier, A., and Lavik, G.: Nitrogen transfers off Walvis https://doi.org/10.1016/j.pocean.2008.10.026, 2008. Bay: a 3-D coupled physical/biogeochemical modeling approach Edwards, A. M. and Yool, A.: The role of higher predation in in the Namibian upwelling system, Biogeosciences, 10, 4117– plankton population models, J. Plankton Res., 22, 1085–1112, 4135, https://doi.org/10.5194/bg-10-4117-2013, 2013b. https://doi.org/10.1093/plankt/22.6.1085, 2000. Hirst, A. G. and Kiørboe, T.: Mortality of marine planktonic cope- Espinoza, P. and Bertrand, A.: Revisiting Peruvian anchovy (En- pods: global rates and patterns, Mar. Ecol. Prog. Ser., 230, 195– graulis ringens) trophodynamics provides a new vision of 209, https://doi.org/10.3354/meps230195, 2002. the Humboldt Current system, Prog. Oceanogr., 79, 215–227, Holbrook, N. J., Brown, J. N., Davidson, J., Feng, M., Hobday, A. J., https://doi.org/10.1016/j.pocean.2008.10.022, 2008. Lough, J. M., McGregor, S., Power, S. B, and Risbey, J. S.: El Evans, G. T. and Parslow, J. S.: A model of annual plankton cy- Niño–Southern Oscillation, in: A Marine Climate Change Im- cles, available at: https://www.tandfonline.com/doi/abs/10.1080/ pacts and Adaptation Report Card for Australia 2012, edited 01965581.1985.10749478 (last access: 23 April 2021), Biologi- by: Poloczanska, E., Hobday, A., and Richardson, A., CSIRO cal Oceanography, 3, 327–347, 1985. Climate Adaptation Flagship, http://hdl.handle.net/102.100.100/ Farías, L., Castro-González, M., Cornejo, M., Charpentier, J., Faún- 99770?index=1 (last access: 23 April 2021), ISBN 978-0-643- dez, J., Boontanon, N., and Yoshida, N.: Denitrification and ni- 10927-8, 2012. trous oxide cycling within the upper oxycline of the eastern tropi- Hsieh, C.-h., Reiss, C. S., Hunter, J. R., Beddington, J. R., cal South Pacific oxygen minimum zone, Limnol. Oceanogr., 54, May, R. M., and Sugihara, G.: Fishing elevates variability 132–144, https://doi.org/10.4319/lo.2009.54.1.0132, 2009. in the abundance of exploited species, Nature, 443, 859–862, Fasham, M. J. R., Ducklow, H. W., and McKelvie, S. M.: https://doi.org/10.1038/nature05232, 2006. A nitrogen-based model of plankton dynamics in Hutchings, L.: Fish harvesting in a variable, produc- tive environment – searching for rules or searching

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS 2915

for exceptions?, S. Afr. J. Marine Sci., 12, 297–318, numerical weather prediction, Geophys. Res. Lett., 25, 761–764, https://doi.org/10.2989/02577619209504708, 1992. https://doi.org/10.1029/98GL00544, 1998. Irigoien, X., Flynn, K. J., and Harris, R. P.: Phytoplankton blooms: Lluch-Belda, D., Crawford, R. J. M., Kawasaki, T., MacCall, a ‘loophole’ in microzooplankton grazing impact?, J. Plankton A. D., Parrish, R. H., Schwartzlose, R. A., and Smith, Res., 27, 313–321, https://doi.org/10.1093/plankt/fbi011, 2005. P. E.: World-wide fluctuations of sardine and anchovy stocks: José, Y. S., Dietze, H., and Oschlies, A.: Linking diverse nutrient the regime problem, S. Afr. J. Marine Sci., 8, 195–205, patterns to different water masses within anticyclonic eddies in https://doi.org/10.2989/02577618909504561, 1989. the upwelling system off Peru, Biogeosciences, 14, 1349–1364, McInturf, A. G., Pollack, L., Yang, L. H., and Spiegel, O.: Vec- https://doi.org/10.5194/bg-14-1349-2017, 2017. tors with autonomy: what distinguishes animal-mediated nutri- Kiko, R. and Hauss, H.: On the Estimation of Zooplankton- ent transport from abiotic vectors?, Biol. Rev., 94, 1761–1773, Mediated Active Fluxes in Oxygen Minimum Zone https://doi.org/10.1111/brv.12525, 2019. Regions, Frontiers in Marine Science, 6, 741, Messié, M. and Chavez, F. P.: Seasonal regulation of primary pro- https://doi.org/10.3389/fmars.2019.00741, 2019. duction in eastern boundary upwelling systems, Prog. Oceanogr., Kiko, R., Brandt, P., Christiansen, S., Faustmann, J., Kriest, I., Ro- 134, 1–18, https://doi.org/10.1016/j.pocean.2014.10.011, 2015. drigues, E., Schütte, F., and Hauss, H.: Zooplankton-Mediated Mitra, A., Castellani, C., Gentleman, W. C., Jónasdóttir, S. H., Fluxes in the Eastern Tropical North Atlantic, Frontiers in Ma- Flynn, K. J., Bode, A., Halsband, C., Kuhn, P., Licandro, rine Science, 7, 358, https://doi.org/10.3389/fmars.2020.00358, P., Agersted, M. D., Agersted, M. D., Calbet, A., Lind- 2020. eque, P. K., Koppelmann, R., Møller, E. F., Gislason, A., Kishi, M. J., Kashiwai, M., Ware, D. M., Megrey, B. A., Eslinger, Nielsen, T. G., and St. John, M.: Bridging the gap be- D. L., Werner, F. E., Noguchi-Aita, M., Azumaya, T., Fujii, tween marine biogeochemical and fisheries sciences; config- M., Hashimoto, S., Huang, D., Iizumi, H., Ishida, Y., Kang, S., uring the zooplankton link, Prog. Oceanogr., 129, 176–199, Kantakov, G. A., Kim, H.-C., Komatsu, K., Navrotsky, V. V., https://doi.org/10.1016/j.pocean.2014.04.025, 2014. Smith, S. L., Tadokoro, K., Tsuda, A., Yamamura, O., Yamanaka, Muck, P., Rojas de Mendiola, B., and Antonietti, E.: Compara- Y., Yokouchi, K., Yoshie, N., Zhang, J., Zuenko, Y. I., and tive studies on feeding in larval anchoveta (Engraulis ringens) Zvalinsky, V. I.: NEMURO – a lower trophic level model for and sardine (Sardinops sagax), in: The Peruvian Upwelling the North Pacific marine ecosystem, Ecol. Model., 202, 12–25, Ecosystem: Dynamics and Interactions. ICLARM Conference https://doi.org/10.1016/j.ecolmodel.2006.08.021, 2007. Proceedings 18, edited by: Pauly, D., Muck, P., Mendo, J., and Kiørboe, T.: Zooplankton body composition, Limnol. Oceanogr., Tsukayama, I., 86–96, Instituto del Mar del Peru (IMARPE), 58, 1843–1850, https://doi.org/10.4319/lo.2013.58.5.1843, Deutsche Gesellschaft für Technische Zusammenarbeit (GTZ), 2013. and International Center for Living Aquatic Resources Manage- Koné, V., Machu, E., Penven, P., Andersen, V., Garçon, V., Fréon, P., ment (ICLARM), Manila, 1989. and Demarcq, H.: Modeling the primary and secondary produc- Mullin, M. M.: An experimentalist’s view of zooplankton models, tions of the southern Benguela upwelling system: A comparative CalCOFI Reports, 18, 132–135, 1975. study through two biogeochemical models, Global Biogeochem. Neubert, M. G., Klanjscek, T., and Caswell, H.: Re- Cy., 19, GB4021, https://doi.org/10.1029/2004GB002427, 2005. activity and transient dynamics of predator–prey Kriest, I., Sauerland, V., Khatiwala, S., Srivastav, A., and Os- and food web models, Ecol. Model., 179, 29–38, chlies, A.: Calibrating a global three-dimensional biogeochemi- https://doi.org/10.1016/j.ecolmodel.2004.05.001, 2004. cal ocean model (MOPS-1.0), Geosci. Model Dev., 10, 127–154, Ohman, M. D.: The Demographic Benefits of Diel Verti- https://doi.org/10.5194/gmd-10-127-2017, 2017. cal Migration by Zooplankton, Ecol. Monogr., 60, 257–281, Lavery, T. J., Roudnew, B., Gill, P., Seymour, J., Seuront, L., https://doi.org/10.2307/1943058, 1990. Johnson, G., Mitchell, J. G., and Smetacek, V.: Iron defeca- Oliveros-Ramos, R., Verley, P., Echevin, V., and Shin, Y.-J.: tion by sperm whales stimulates carbon export in the South- A sequential approach to calibrate ecosystem models with ern Ocean, P. Roy. Soc. B-Biol. Sci., 277(1699), 3527–3531, multiple time series data, Prog. Oceanogr., 151, 227–244, https://doi.org/10.1098/rspb.2010.0863, 2010. https://doi.org/10.1016/j.pocean.2017.01.002, 2017. Lehette, P. and Hernández-León, S.: Zooplankton biomass estima- O’Reilly, J. E., Maritorena, S., Mitchell, B. G., Siegel, D. A., tion from digitized images: a comparison between subtropical Carder, K. L., Garver, S. A., Kahru, M., and McClain, C.: Ocean and Antarctic organisms, Limnol. Oceanogr.-Meth., 7, 304–308, color chlorophyll algorithms for SeaWiFS, J. Geophys. Res.- https://doi.org/10.4319/lom.2009.7.304, 2009. Oceans, 103, 24937–24953, https://doi.org/10.1029/98JC02160, Lima, I. D. and Doney, S. C.: A three-dimensional, mult- 1998. inutrient, and size-structured ecosystem model for the Pauly, D., Jarre, A., Luna, S., Sambilay, V., Rojas de Mendiola, North Atlantic, Global Biogeochem. Cy., 18, GB3019, J. B., and Alamo, A.: On the quantity and types of food ingested https://doi.org/10.1029/2003GB002146, 2004. by Peruvian anchoveta, 1953–1982, in: The Peruvian Upwelling Lima, I. D., Olson, D. B., and Doney, S. C.: Intrin- Ecosystem: Dynamics and Interactions. ICLARM Conference sic dynamics and stability properties of size-structured Proceedings 18, edited by: Pauly, D., Muck, P., Mendo, J., and pelagic ecosystem models, J. Plankton Res., 24, 533–556, Tsukayama, I., 109–124, Instituto del Mar del Peru (IMARPE), https://doi.org/10.1093/plankt/24.6.533, 2002. Deutsche Gesellschaft für Technische Zusammenarbeit (GTZ), Liu, W. T., Tang, W., and Polito, P. S.: NASA scatterometer pro- and International Center for Living Aquatic Resources Manage- vides global ocean-surface wind fields with more structures than ment (ICLARM), Manila, 1989.

https://doi.org/10.5194/bg-18-2891-2021 Biogeosciences, 18, 2891–2916, 2021 2916 M. Hill Cruz et al.: Zooplankton mortality effects on the plankton community of the NHCS

Prowe, A. E. F., Pahlow, M., Dutkiewicz, S., Follows, M., and Tian, R. C., Vézina, A. F., Starr, M., and Saucier, F.: Seasonal dy- Oschlies, A.: Top-down control of marine phytoplankton diver- namics of coastal ecosystems and export production at high lat- sity in a global ecosystem model, Prog. Oceanogr., 101, 1–13, itudes: a modeling study, Limnol. Oceanogr., 46, 1845–1859, https://doi.org/10.1016/j.pocean.2011.11.016, 2012. https://doi.org/10.4319/lo.2001.46.8.1845, 2001. Remsen, A., Hopkins, T. L., and Samson, S.: What you see Tittensor, D. P., Eddy, T. D., Lotze, H. K., Galbraith, E. D., Cheung, is not what you catch: a comparison of concurrently col- W., Barange, M., Blanchard, J. L., Bopp, L., Bryndum-Buchholz, lected net, Optical Plankton Counter, and Shadowed Image A., Büchner, M., Bulman, C., Carozza, D. A., Christensen, V., Particle Profiling Evaluation Recorder data from the north- Coll, M., Dunne, J. P., Fernandes, J. A., Fulton, E. A., Hobday, east Gulf of Mexico, Deep-Sea Res. Pt. I, 51, 129–151, A. J., Huber, V., Jennings, S., Jones, M., Lehodey, P., Link, J. https://doi.org/10.1016/j.dsr.2003.09.008, 2004. S., Mackinson, S., Maury, O., Niiranen, S., Oliveros-Ramos, R., Ridgway K. R., Dunn, J. R., and Wilkin, J. L.: Ocean In- Roy, T., Schewe, J., Shin, Y.-J., Silva, T., Stock, C. A., Steenbeek, terpolation by Four-Dimensional Weighted Least Squares J., Underwood, P. J., Volkholz, J., Watson, J. R., and Walker, – Application to the Waters around Australia, J. Atmos. N. D.: A protocol for the intercomparison of marine fishery Ocean. Tech., 19, 1357–1375, https://doi.org/10.1175/1520- and ecosystem models: Fish-MIP v1.0, Geosci. Model Dev., 11, 0426(2002)019<1357:OIBFDW>2.0.CO;2, 2002. 1421–1442, https://doi.org/10.5194/gmd-11-1421-2018, 2018. Rose, K. A., Allen, J. I., Artioli, Y., Barange, M., Blackford, J., Travers-Trolet, M., Shin, Y.-J., and Field, J.: An end-to-end Carlotti, F., Cropp, R., Daewel, U., Edwards, K., Flynn, K., Hill, coupled model ROMS-N2P2Z2D2-OSMOSE of the south- S. L., HilleRisLambers, R., Huse, G., Mackinson, S., Megrey, ern Benguela foodweb: parameterisation, calibration and B., Moll, A., Rivkin, R., Salihoglu, B., Schrum, C., Shannon, pattern-oriented validation, Afr. J. Mar. Sci., 36, 11–29, L., Shin, Y.-J., Smith, S. L., Smith, C., Solidoro, C., St. John, https://doi.org/10.2989/1814232X.2014.883326, 2014a. M., and Zhou, M.: End-To-End Models for the Analysis of Ma- Travers-Trolet, M., Shin, Y.-J., Shannon, L. J., Moloney, C. L., rine Ecosystems: Challenges, Issues, and Next Steps, Mar. Coast. and Field, J. G.: Combined Fishing and Climate Forcing in Fish., 2, 115–130, https://doi.org/10.1577/C09-059.1, 2010. the Southern Benguela Upwelling Ecosystem: An End-to-End Rykaczewski, R. R. and Checkley, D. M.: Influence of ocean winds Modelling Approach Reveals Dampened Effects, PloS one, 9, on the pelagic ecosystem in upwelling regions, Proc. Natl. Acad. e94 286, https://doi.org/10.1371/journal.pone.0094286, 2014b. Sci., 105, 1965–1970, https://doi.org/10.1073/pnas.0711777105, van der Lingen, C. D., Hutchings, L., and Field, J. G.: Com- 2008. parative trophodynamics of anchovy Engraulis encrasico- Ryther, J. H.: Photosynthesis and fish production in the sea, Science, lus and sardine Sardinops sagax in the southern Benguela: 166, 72–76, 1969. are species alternations between small pelagic fish tropho- Shchepetkin, A. F. and McWilliams, J. C.: The regional oceanic dynamically mediated?, Afr. J. Mar. Sci., 28, 465–477, modeling system (ROMS): a split-explicit, free-surface, https://doi.org/10.2989/18142320609504199, 2006. topography-following-coordinate oceanic model, Ocean Model., Varpe, Ø., Fiksen, Ø., and Slotte, A.: Meta-ecosystems and bi- 9, 347–404, https://doi.org/10.1016/j.ocemod.2004.08.002, ological energy transport from ocean to coast: the ecologi- 2005. cal importance of herring migration, Oecologia, 146, 443–451, Shimoda, Y. and Arhonditsis, G. B.: Phytoplankton functional type https://doi.org/10.1007/s00442-005-0219-9, 2005. modelling: Running before we can walk? A critical evaluation Williams, J. J., Papastamatiou, Y. P., Caselle, J. E., Bradley, of the current state of knowledge, Ecol. Model., 320, 29–43, D., and Jacoby, D. M. P.: Mobile marine predators: an https://doi.org/10.1016/j.ecolmodel.2015.08.029, 2016. understudied source of nutrients to coral reefs in an un- Steele, J. H. and Henderson, E. W.: The role of preda- fished atoll, P. Roy. Soc. B-Biol. Sci., 285, 20172456, tion in plankton models, J. Plankton Res., 14, 157–172, https://doi.org/10.1098/rspb.2017.2456, 2018. https://doi.org/10.1093/plankt/14.1.157, 1992. Worley, S. J., Woodruff, S. D., Reynolds, R. W., Lubker, S. J., and Suntharalingam, P., Sarmiento, J. L., and Toggweiler, Lott, N.: ICOADS release 2.1 data and products, Int. J. Climatol., J. R.: Global significance of nitrous-oxide production 25, 823–842, https://doi.org/10.1002/joc.1166, 2005. and transport from oceanic low-oxygen zones: A mod- Yakushev, E. V., Pollehne, F., Jost, G., Kuznetsov, I., Schnei- eling study, Global Biogeochem. Cy., 14, 1353–1370, der, B., and Umlauf, L.: Analysis of the water col- https://doi.org/10.1029/1999GB900100, 2000. umn oxic/anoxic interface in the Black and Baltic seas Suntharalingam, P., Buitenhuis, E., Le Quéré, C., Dentener, F., with a numerical model, Mar. Chem., 107, 388–410, Nevison, C., Butler, J. H., Bange, H. W., and Forster, G.: https://doi.org/10.1016/j.marchem.2007.06.003, 2007. Quantifying the impact of anthropogenic nitrogen deposition on oceanic nitrous oxide, Geophys. Res. Lett., 39, L07605, https://doi.org/10.1029/2011GL050778, 2012. Tian, R. C., Vézina, A., Legendre, L., Ingram, R. G, Klein, B., Packard, T., Roy, S., Savenkoff, C., Silverberg, N., Therriault, J. C., and Tremblay, J. E.: Effects of pelagic food-web in- teractions and nutrient remineralization on the biogeochemi- cal cycling of carbon: a modeling approach, Deep-Sea Res. Pt. II, 47, 637–662, https://doi.org/10.1016/S0967-0645(99)00121- 6, 2000.

Biogeosciences, 18, 2891–2916, 2021 https://doi.org/10.5194/bg-18-2891-2021